J. Anim Sci.
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

J. Anim Sci. 2007. 85:3218-3227. doi:10.2527/jas.2007-0332
© 2007 American Society of Animal Science
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow All Versions of this Article:
jas.2007-0332v1
85/12/3218    most recent
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Kause, A.
Right arrow Articles by Koskinen, H.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Kause, A.
Right arrow Articles by Koskinen, H.

ANIMAL GENETICS

Direct and indirect selection of visceral lipid weight, fillet weight, and fillet percentage in a rainbow trout breeding program1

A. Kause*,2, T. Paananen{dagger}, O. Ritola{dagger} and H. Koskinen{dagger}

* MTT Agrifood Research Finland, Biotechnology and Food Research, Biometrical Genetics, FI-31600 Jokioinen, Finland; and {dagger} Finnish Game and Fisheries Research Institute, Tervo Fisheries Research and Aquaculture, FI-72210 Tervo, Finland


   Abstract
Top
 Abstract
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 LITERATURE CITED
 
We assessed whether visceral lipid weight, fillet weight, and percentage fillet from BW, 3 traits laborious to record, could be genetically improved by indirect selection on more easily measured traits in farmed rainbow trout. Visceral lipid is discarded as waste during slaughter, influencing production efficiency and production costs. Fillet weight and fillet percentage directly influence economic returns in trout production. The study comprised 3 steps. First, we assessed the degree to which selection on percentage of visceral weight from BW indirectly changes visceral lipid weight and the size of intestines and internal organs. The phenotypic analysis of weights of viscera, intestines, visceral lipid, liver, and gonads measured from 40 fish revealed that phenotypic selection against visceral weight was most strongly directed to visceral lipid, and to a lesser degree to intestines and gonads. Because genetic relationships among these traits were not established, it is not known whether indirect selection leads to genetic responses. Second, we examined whether direct selection for the fillet traits could be replaced by indirect selection on BW, eviscerated BW, visceral weight, visceral percentage, head volume, and relative head volume (head volume relative to BW). The selection index calculations based on the quantitative genetic parameters obtained from multigenerational pedigree data showed that genetic improvement of fillet percentage through direct selection (selection accuracy, rTI = 0.54) was equally efficient compared with indirect selection on visceral percentage ( rTI = 0.54). Genetic improvement of fillet weight through direct selection (rTI = 0.56) was always more efficient than indirect selection, yet indirect selection for eviscerated BW ( rTI = 0.50) was almost as efficient as direct selection. Third, the expected genetic responses to alternative selection indices showed that improved fillet percentage was mainly a result of a moderate decrease in visceral weight rather than of a major increase in absolute fillet weight. Moreover, fillet percentage is challenging to improve, even if it exhibits moderate heritability (h2 = 0.29). This is because fillet percentage displays low phenotypic variation. In conclusion, fillet weight and fillet percentage can be increased by indirect selection against visceral percentage and for high eviscerated BW.

Key Words: fillet yield • genetic correlation • heritability • lipid deposition • quantitative genetics • rainbow trout


   INTRODUCTION
Top
 Abstract
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 LITERATURE CITED
 
We assessed whether visceral lipid, fillet weight, and fillet percentage (percentage of fillet from BW), 3 traits laborious to record from thousands of individuals, could be genetically improved by indirect selection on more easily measured traits in farmed rainbow trout Oncorhynchus mykiss (Walbaum). Visceral lipid is discarded during slaughter, whereas fillet weight and fillet percentage directly influence economic returns in trout production.

Promisingly, visceral percentage (percentage dressing waste from BW, or its converse, dressing percent) is favorably genetically correlated with visceral lipid (Gjerde and Schaeffer, 1989Go; Kause et al., 2002Go). This could provide the possibility of using visceral percentage, an easily measured trait, as an indirect measure of visceral lipid. However, this poses a potential risk. Selection against visceral weight may reduce the relative size of internal organs and intestines that are fundamental for fish physiology (Bergot et al., 1981Go; Poppe et al., 2003Go).

Previous studies have examined the potential for predicting fillet weight and fillet percentage of fish from body dimensions, ultrasound scans, and head size (Bosworth et al., 1998Go, 2001Go; Cibert et al., 1999Go; Rutten et al., 2004Go). These studies can be extended by assessing whether selection against visceral weight would simultaneously decrease visceral lipid and increase fillet percentage.

First, we assessed how selection on visceral weight would indirectly select on the weight of visceral lipid, intestines, liver, and gonads. Second, to examine whether direct selection for the fillet traits could be replaced by indirect selection on BW, eviscerated BW, visceral weight, visceral percentage, head volume, and relative head volume (head volume relative to BW), heritabilities and phenotypic and genetic correlations for these traits were estimated. Third, by using selection index calculations, the accuracy of alternative selection indices in predicting fillet percentage and fillet weight was calculated.


   MATERIALS AND METHODS
Top
 Abstract
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 LITERATURE CITED
 
Fish management was approved by the animal care and use committee of the Finnish Game and Fisheries Research Institute.

In the Finnish breeding program for rainbow trout, breeding candidates are held at the fresh water nucleus. Their sibs are performance tested in the sea under commercial production conditions and slaughtered to record product quality traits (Kause et al., 2005Go). This study was conducted to evaluate the value of recording new traits from these fish for the improvement of fillet traits and visceral lipid.

Population Structure

The fish studied originated from the Finnish national breeding program. The broodstock management, selection, and mating procedures have been described by Kause et al. (2005)Go. The fresh water nucleus breeding station is located in Tervo in central Finland (latitude: 60° 1' 26''; longitude: 26° 39' 40'').

Each year in April, approximately 300 families are produced at the nucleus station. The matings have been either paternal nested or partial factorial designs, and from year 2003 onward each sire and dam vary in the number of partners, as determined by the method of optimal genetic contributions (Wray and Goddard, 1994Go). At the eyed-egg stage in June, each family is transferred to a 150-L indoor tank. The families are held separated until tagging, from November to January, at a weight of approximately 50 g. The fish are tagged by using passive integrated transponders (Trovan, Köln, Germany). At tagging, each family is split into 2 or 3 groups to be reared at the fresh water nucleus station, and at 1 or 2 sea test stations (Kause et al., 2005Go). To increase genetic gain at the nucleus station, from year 2003 onward within-family selection has been practiced during tagging by leaving the largest fish within a family at the nucleus and sending the second largest ones to the sea stations (Martinez et al., 2006Go). The remaining untagged fish within a family are weighed as a group (w), counted (n), and culled.

At a sea station, the fish are held in a single net cage and managed following the commercial practices of the farm. After one sea-growing season, from April to the next winter, the fish are slaughtered at a weight of approximately 1 to 2 kg. All fish analyzed for this study were reared at the sea water stations until harvest.

Small Data Set: Data Collection

To assess phenotypic relationships between visceral weight and its components, a total of 40 fish from generation 2003 were randomly sampled during slaughter at the sea station located at Åland Islands (latitude: 60° 2' 33''; longitude: 19° 57' 27''). A small number of fish were used because it was not realistic to dissect and weigh the components of visceral weight from a large number of fish. In salmonids, visceral lipid is tightly attached to the internal organs. Accordingly, by using this data, we can only assess how direct selection on visceral weight indirectly selects the component traits. Thus, we are unable to predict correlated genetic responses in the component traits.

All sampled fish were immature and were identified as being males (n = 11), females (n = 20), or fish of unknown sex (n = 9) based on the investigation of gonads. The average fish weight was 1,574 ± 270 g (±SD). The fish were recorded for 5 traits. First, the fish were dressed and visceral weight (158 ± 31.8 g) was measured. Thereafter, 4 components of viscera, namely, intestines (78.6 ± 15.9 g), visceral lipid (56.2 ± 22.3 g), liver (18.7 ± 4.56 g), and gonads (0.869 ± 0.414 g), were dissected and weighed.

Small Data Set: Statistical Analysis

Selection index calculations were used to assess the strength of indirect selection on intestine, visceral lipid, liver, and gonad weights resulting from direct selection on visceral weight. First, the phenotypic variance-covariance matrix among the 5 traits (measured in grams) was estimated. Visceral weight was then assumed to be selected downward by 1 SD (i.e., with a selection intensity of 1), and selection differential (Svisce) was calculated for visceral weight. Selection differential is the difference in a trait mean between the whole population and the selected group. Here, selection differentials are also expressed as the percentage from a population mean of a trait (S%). Selection differential of a component trait x, resulting from direct selection against visceral weight, was calculated as:


Formula

where b equals the regression coefficient between the component trait x and visceral weight (Falconer and Mackay, 1996Go, p. 317). Before the calculation of selection differentials, the effect of sex was removed from the data by running a 1-way ANOVA with sex as the fixed factor for each trait, and by using the residuals of the models in the successive analysis.

Large Data Set: Data Collection

The large data set used to estimate genetic parameters included 29,666 fish with observations recorded at the sea station (Table 1Go). The fish originated from 1 of 6 generations (1998 to 2004 but excluding 2002), and each generation was reared at 1 or 2 sea test stations. The population structure is given in Table 1Go.


View this table:
[in this window]
[in a new window]

  		Table 1. Numbers of observations, generations, sea station x generation classes (stat x gen), sires, dams, full-sib families, and family size in the large data set
 
All fish were measured for BW and then eviscerated, and eviscerated BW was recorded. Based on visual inspection of gonads, the fish were classified into mature males, immature males, immature females, and fish of unknown sex. Fillet weight of randomly sampled fish was recorded in generations 2003 and 2004 at 2 (n = 1953 fish) and 1 sea station(s) (n = 718 fish), respectively. To record untrimmed fillet weight (with skin, ribs, and lipid deposits included), fish were filleted from one side of the body and the weight of the fillet was multiplied by 2. Head dimensions were recorded in generations 2003 (n = 929) and 2004 (n = 669) at 1 sea station, with the same fish as for the fillet weight recording. Head length (from the snout to the back of the operculum), head height (behind the operculum), and head width (behind an eye) were measured to the nearest 1 mm by using a pair of calipers. Head volume was calculated as: head volume = [(head length x head height)/2] x head width. This formula assumes that the head is triangle shaped and evenly wide.

Eight traits were analyzed: BW, eviscerated BW, visceral weight (BW – eviscerated BW), visceral percentage [100 x (visceral weight/BW)], fillet weight, fillet percentage [100 x (fillet weight/BW)], head volume, and relative head volume (head volume/BW). The sample sizes are given in Table 1Go. Although eviscerated BW and visceral weight are statistically independent, visceral percentage is the percentage of BW remaining after subtracting the percentage of eviscerated BW [or dressing percent: 100 x (eviscerated BW/BW)]. Thus, visceral percentage and eviscerated BW percentage exhibit the same genetic parameters, yet their correlations are of different signs.

Large Data Set: Genetic Analysis

Phenotypic (rP) and genetic correlations (rG) and heritabilities (h2) were estimated by using the DMUAI software (Jensen and Madsen, 2000Go). The pedigree was traced back to 1989 for use in the analysis. The multitrait animal model used was:


Formula

where animi is a random genetic effect of a individual i; fy x fseffectj is a random fertilization year x full-sib family interaction; FY x MAT x SEX x STATk is a fixed effect for the interaction of fertilization year, maturity, sex, and sea test station; and eijk is a random residual of an observation yijk.

The preselection of fish within families was accounted for in the genetic analysis to obtain unbiased variance components (Ouweltjes et al., 1988Go). This was done by always including the tagging weight of all fish (n = 137,071 fish, including those sent to the sea, those left in the nucleus, and those culled), as one of the traits in all multitrait analyses. Genetic parameters and the statistical model for tagging weight have been presented by Kause et al. (2005)Go and are not repeated here.

Heritabilities (h2 = VG/VP) and full-sib effect ratios (c2 = VFS/VP) were calculated, where VG is the genetic variance attributable to the animal effect, VFS is the variance attributable to fertilization year x full-sib family interaction, and VP is phenotypic variance. The full-sib effect was modeled without pedigree information, and it includes effects attributable to common rearing of full sibs from incubation until tagging, as well as parts of potential dominance and maternal effects. For visceral percentage, fillet percentage, and relative head volume, the full-sib effect was negligible (c2 ≤ 0.02) and was excluded from subsequent analyses.

Several multitrait analyses were needed to obtain all correlations among the traits. Thereafter, the full phenotypic and genetic correlation matrices were constructed, and they were bent to be positive definite by using the method of Hayes and Hill (1981)Go. In particular, bending reduced those correlations close to unity, with maximum changes in phenotypic and genetic correlations of 0.02 and 0.05, respectively.

Large Data Set: Selection Index Calculations

Selection index theory was used to calculate the accuracy (rTI) of alternative selection indices used for the direct and indirect improvement of either fillet percentage or fillet weight (Hazel, 1943Go; Cameron, 1997Go, p. 64). Selection was assumed to be mass selection. Index weights were set to maximize the correlation between the index and the breeding objective. This approach is only indicative because the real breeding program is based on breeding value evaluations (Kause et al., 2005Go). Moreover, in reality, selection on eviscerated BW, fillet, and visceral traits is necessarily sib selection, whereas BW and head volume can also be measured on the live breeding candidates. To assess the consequences of the alternative selection strategies, the expected genetic gains in response to different selection indices were calculated (Cameron, 1997Go). Accuracies and genetic responses were calculated separately for selection strategies in which either fillet percentage or fillet weight was the breeding goal.


   RESULTS
Top
 Abstract
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
DISCUSSION
 LITERATURE CITED
 
Small Data Set: Selection Differentials

Selection index calculations for the phenotypic data of 40 fish showed that selection against visceral weight with a selection intensity of 1 resulted in all component weight traits being indirectly selected downward (Table 2Go). Selection for visceral weight resulted in a 19.7% difference in visceral weight between the whole population and the selected group (i.e., selection differential expressed as the percentage change). As a result of indirect selection, the selection differential percentage for visceral lipid was 31.5%. This was more than for the directly selected visceral weight. Selection differential percentages for all the other component weight traits were less or of a similar amount compared with visceral weight, with values ranging from 6.13 to 19.9%. Consequently, the visceral lipid percentage of the selected group was lower than the percentage in the whole population (a decrease from 35.5 to 30.4%), whereas percentages of intestines and gonads were increased in the selected group. The percentage of liver weight remained very stable between the whole population and the selected group (Table 2Go). These results show that selection on visceral weight was most strongly directed to visceral lipid weight, and that visceral lipid weight was the only component trait whose proportion was decreasing when selecting against visceral weight.


View this table:
[in this window]
[in a new window]

  		Table 2. Selection differential expressed in grams (S) and in percentage (S%) from a population mean resulting from directional selection against visceral weight with a selection intensity of 1
 
Large Data Set: Heritabilities of Traits

Heritabilities for all weight- and volume-based traits were moderate (0.29 to 0.35; Table 3Go). Likewise, heritabilities for fillet percentage (0.29) and relative head volume (0.23) were moderate. The highest heritabilities were found for visceral weight (0.35) and visceral percentage (0.58). The highest CV was found for visceral weight and the lowest for fillet percentage (Table 3Go).


View this table:
[in this window]
[in a new window]

  		Table 3. Mean, phenotypic variance (VP), CV (CVP), heritability (h2), full-sib effect ratio (c2), and their SE for traits in the large data set
 
Large Data Set: Correlations of Weight-Based and Volume-Based Traits

As expected, phenotypic and genetic correlations between weight and eviscerated BW were very strong (Table 4Go). All correlations of BW and eviscerated BW with fillet weight were equal to or higher than 0.93, indicating severe constraints on increasing fillet weight and BW independently. In contrast, phenotypic and genetic correlations of visceral weight and head volume with BW and eviscerated BW were clearly lower than unity (0.70 to 0.85; Table 4Go), indicating that visceral weight and head volume can be more freely changed independently of BW.


View this table:
[in this window]
[in a new window]

  		Table 4. Phenotypic (above the diagonal) and genetic correlations (below the diagonal; ± SE) between traits defined as weights or volume in the large data set
 
Large Data Set: Correlations Among BW and Percentage Traits

Phenotypic and genetic correlations of BW with visceral percentage were weakly positive and unfavorable (0.16 to 0.19; Table 5Go). For eviscerated BW, the respective correlations were weaker (0.05 to 0.11). This result revealed a slightly more favorable correlation structure for eviscerated BW compared with BW.


View this table:
[in this window]
[in a new window]

  		Table 5. Phenotypic (above the diagonal) and genetic correlations (below the diagonal; ±SE) between BW, eviscerated BW, and traits defined as proportions of BW in the large data set
 
Phenotypic correlations of fillet percentage with BW and eviscerated BW were positive and favorable (Table 5Go). However, the genetic correlation between fillet percentage and BW was weak. The genetic correlation of fillet percentage with eviscerated BW was weakly positive and significant, and thus more favorable than with BW. Phenotypic correlations of relative head volume with BW and eviscerated BW were moderately negative and favorable, but the respective genetic correlations were only weakly negative (Table 5Go).

Large Data Set: Correlations Among Percentage Traits

Phenotypic and genetic correlations between fillet percentage and visceral percentage were strongly negative, which is favorable and implies that fillet percentage can be indirectly improved by decreasing visceral percentage (Table 5Go). For relative head volume, phenotypic correlation with fillet percentage was moderately negative and the genetic correlation was negative but low. This revealed a low potential for using relative head volume to select indirectly for improved fillet percentage.

Phenotypic and genetic correlations between visceral percentage and relative head volume were negative and moderate (Table 5Go). Because both visceral and head percentages contribute to production efficiency, this relationship can be regarded as unfavorable.

Large Data Set: Selection for Fillet Percentage

Accuracy (rTI) for direct selection of fillet percentage was 0.54 (Table 6Go, direct selection). The accuracy was increased to 0.63 when visceral percentage (V%) was included in the index (Table 6Go, combined direct and indirect selection). Including eviscerated BW (EBW) or relative head volume along with fillet percentage did not significantly increase accuracy (Table 6Go, combined direct and indirect selection).


View this table:
[in this window]
[in a new window]

  		Table 6. Accuracy (rTI) of alternative selection indices for improving fillet percentage1
 
Indirect selection for fillet percentage was an effective alternative to direct selection (Table 6Go, indirect selection). Selection against visceral percentage (rTI = 0.54) was equally effective compared with direct selection. Accuracy for relative head volume was low. The use of indices I (V%,Hrel) and I (EBW,V%,Hrel) increased the accuracy to 0.59 and 0.60, respectively (Table 6Go, indirect selection).

Expected genetic responses for a set of indices to select for fillet percentage (F%) are given in Table 7Go (maximize fillet percentage). Direct selection for fillet percentage resulted in a modest genetic change in fillet percentage. Selection for index I (F%) increased the fillet percentage by 1.11%. This increase was a result of a moderate decrease in absolute visceral weight (–3.10%) and a smaller increase in absolute fillet weight (1.95%). Relative head volume was changed only slightly. This phenomenon was exaggerated when using index I (F%,V%,Hrel) and indirect selection index I (V%,Hrel). In these cases, visceral weight was reduced by 6.2 to 7.0% and fillet weight was increased by only 1.6 to 1.9%. This result occurred because fillet percentage (and fillet weight) displayed lower heritability and lower phenotypic variation than visceral percentage (and visceral weight; Table 3Go). Index I (F%,V%,Hrel) produced the maximum genetic gain of 1.32% in fillet percentage (Table 7Go, maximize fillet percentage), which means that fillet percentage was increased from 64.75 to 65.60%.


View this table:
[in this window]
[in a new window]

  		Table 7. Expected genetic gains (in units of percentage from original mean) in response to 6 selection indices (I)1
 
Large Data Set: Selection for Fillet Weight

Direct selection for fillet weight resulted in an accuracy of 0.56 (Table 8Go, direct selection). Adding combinations of visceral percentage, relative head volume, eviscerated BW, and fillet percentage to the index along with fillet weight increased the accuracy only marginally, and the maximum accuracy was 0.59 (Table 8Go, combined direct and indirect selection).


View this table:
[in this window]
[in a new window]

  		Table 8. Accuracy (rTI) of alternative selection indices for improving fillet weight1
 
Indirect selection for fillet weight was also effective, but was always less effective than direct selection (Table 8Go, indirect selection). The accuracy of the index that included eviscerated BW (rTI = 0.50) was almost as high as for direct selection. Accuracy was slightly higher with index I (EBW,V%) (0.52). Adding relative head volume, visceral weight, or head volume to these indices did not increase the accuracy (Table 8Go, indirect selection).

Expected genetic gains in fillet weight are given in Table 7Go (maximize fillet weight). Selection for fillet weight resulted in 6.43% genetic gain. Simultaneously, fillet percentage was improved by 0.36% (from 64.75 to 64.98%). Selection for eviscerated BW resulted in a 5.80% gain in fillet weight and fillet percentage was also improved, but visceral percentage increased. Selection for index I (EBW,V%) led to a 6.03% gain in fillet weight, a 0.448% increase in fillet percentage (from 64.75% to 65.04), and a 2.02% decrease in visceral percentage (Table 7Go, maximize fillet weight). Hence, the use of index I (EBW,V%) resulted in a major genetic increase in fillet weight, a slight increase in fillet percentage, and simultaneously controlled for genetic changes occurring in visceral percentage.


   DISCUSSION
Top
 Abstract
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
LITERATURE CITED
 
This study produced 4 major results that can be used to improve rainbow trout breeding programs. First, data from 40 fish showed that selection on visceral percentage is an effective indirect way to select on visceral lipid. However, selection for reduced visceral percentage will also select for a reduced size of internal organs. Because of the small amount of phenotypic data used, correlated genetic responses could not be predicted. Second, the genetic analysis of the large pedigree data showed that fillet weight and fillet percentage can be genetically improved by indirect selection on easily measured slaughter traits, such as eviscerated BW and visceral percentage. Third, genetic improvement in fillet percentage results mostly from reduced visceral weight rather than increased fillet weight. Fourth, improving the fillet percentage is challenging, because, despite the moderate heritability, fillet percentage displays low phenotypic variation.

Relation of Visceral Weight to Its Component Traits

The results showed that visceral lipid weight is the component of visceral weight to which selection against or for visceral weight is most strongly directed. It should be noted, however, that selection against visceral weight is associated with indirect selection against weight of all the component traits. Reducing the size or changing the shape of internal organs and intestines may have detrimental effects on fish health and biological efficiency (e.g., Bergot et al., 1981Go; Poppe et al., 2003Go). Likewise, in salmonids lipid stores are important for life functions such as reproduction (Shearer, 1994Go). Thus, it is advisable to monitor whether selection on visceral percentage has a negative impact on health or efficiency. Because selection for rapid growth tends to increase the body lipid percentage as a correlated genetic response (Gjedrem, 1997Go; Kause et al., 2007Go), visceral percentage should be selected to maintain at least a stable visceral percentage.

Previous genetic studies on percentage of eviscerated BW (the percentage of BW remaining after subtraction of visceral percentage) and visceral lipid are in line with our results. Gjerde and Schaeffer (1989Go; rP = –0.39, rG = –0.68) and Kause et al. (2002Go; rP = –0.26, rG = –0.57) found negative phenotypic and genetic correlations between percentage of eviscerated BW and the visceral lipid score in rainbow trout. Similarly, Neira et al. (2004Go; rP = –0.04, rG = –0.28), working on Coho salmon, and Rye and Gjerde (1996Go; rP = –0.24, rG = –0.64), working on Atlantic salmon, found negative correlations between percentage of eviscerated BW and percentage of visceral lipid weight from BW. These results suggest that our observation that decreasing visceral lipid is correlated with decreasing visceral weight should also be a genetically based relationship. Visceral percentage displays positive, yet only moderate, genetic correlations with muscle and whole body lipid percentages (Tobin et al., 2006Go), implying that selection for reduced visceral percentage may also reduce the other lipid traits.

We did not estimate genetic parameters for the components of visceral weight, but some have been reported in the literature. Heritabilities for intestinal weight, visceral lipid weight, gonad weight, visceral fat score, liver weight, and egg volume have ranged from 0.20 to 0.74 (Gall, 1975Go; Gjerde and Schaeffer, 1989Go; Rye and Gjerde, 1996Go; Su et al., 1997Go; Elvingson and Johansson, 1993Go; Kause et al., 2002Go; Neira et al., 2004Go). Consequently, these traits are indeed expected to display genetic responses to selection.

Heritabilities for BW and Viscera

Heritabilities for BW and eviscerated BW were the same (0.29 vs. 0.29). This was expected, because previously estimated heritabilities for BW and eviscerated BW have been 0.31 and 0.31 (Gjerde and Schaeffer, 1989Go), 0.35 and 0.29 (Iwamoto et al., 1990Go), 0.50 and 0.48 (Elvingson and Johansson, 1993Go), 0.45 and 0.43 (Elvingson and Nilsson, 1994Go), 0.20 and 0.22 (Kause et al., 2002Go), and 0.19 and 0.17 (Neira et al., 2004Go).

For visceral percentage we found higher heritability (h2 = 0.58) than for BW and eviscerated BW. This is in the upper range of the previous estimates for percentage of eviscerated BW (0.19 to 0.45; Gjerde and Schaeffer, 1989Go; Rye and Gjerde, 1996Go; Kause et al., 2002Go; Neira et al., 2004Go). In our data, heritability for visceral weight was moderate (h2 = 0.35), and phenotypic and genetic correlations between visceral weight and BW were lower than unity, implying genetic potential for changing their proportional relations.

Relationships Between BW and Viscera

Phenotypic and genetic correlations between BW and eviscerated BW were very strong (r ≥ 0.94), showing that when selecting solely for growth, either BW or eviscerated BW can be equally included in a selection index. This is consistent with other studies (Gjerde and Schaeffer, 1989Go; Iwamoto et al., 1990Go; Elvingson and Nilsson, 1994Go; Kause et al., 2002Go; Neira et al., 2004Go).

An interesting finding was that eviscerated BW displayed slightly less unfavorable phenotypic and genetic correlations with visceral percentage compared with BW. This was expected, because when visceral weight is increasing, it contributes to increasing the BW, leading to a positive correlation between BW and visceral percentage. In contrast, visceral weight does not contribute to eviscerated BW; thus, increasing the visceral weight does not automatically lead to a positive correlation between eviscerated BW and percentage of viscera from BW. This result is also supported by other studies. In rainbow trout, visceral lipid score was more strongly correlated with BW (rP = 0.27, rG = 0.38) than with eviscerated BW (rP = 0.24, rG = 0.22), and percentage of eviscerated BW was more favorably correlated with eviscerated BW ( rP = –0.07, rG = 0.24) than with BW (rP = –0.20, rG = 0.04; Kause et al., 2002Go). These studies imply that when simultaneously breeding for growth and against visceral lipid (or visceral percentage), the use of eviscerated BW in a selection index is slightly more advantageous than the use of BW. This practice is also justified because in many instances, as in Finland, eviscerated BW is the trait of economic interest.

Selection Strategy to Reduce Visceral Lipid by Using Visceral Percentage

The suggested strategy to control for genetic changes in visceral lipid is to select on visceral percentage. Selection on visceral weight was most strongly directed to visceral lipid weight, and to a lesser degree to the intestines, gonads, and liver. This pattern is beneficial for breeders wanting to change the lipid component of visceral weight while minimizing changes in the weight of internal organs. It is not known, however, how much visceral percentage can be reduced without influencing fish health.

Heritabilities for Fillet Traits and Head Volume

Heritabilities for fillet weight (0.31) and fillet percentage (0.29) were moderate and of the same magnitude compared with heritabilities for BW and eviscerated BW. These estimates are in line with previous studies. For fillet percentage, previously estimated heritabilities were 0.33 in rainbow trout (Kause et al., 2002Go), 0.11 in Coho salmon (Neira et al., 2004Go), 0.12 in Nile tilapia (Rutten et al., 2005Go), and 0.38 in common carp (Kocour et al., 2007Go). In our study, heritabilities for both absolute head volume (0.35) and relative head volume (0.23) were moderate. Rutten et al. (2005)Go estimated heritabilities of 0.15 and 0.12 for head weight and percentage of head weight, respectively, in Nile tilapia. Kocour et al. (2007)Go estimated moderate heritabilities (0.15 to 0.32) for relative head length, height, and width in common carp. Head dimensions thus exhibit genetic characteristics similar to other morphological traits. Generally, results from our study and those of others indicate moderate genetic potential for these types of traits.

Although fillet percentage displays moderate heritability, its phenotypic variation is low. This results in a low selection differential for fillet percentage, constraining its improvement. Coefficients of variation for percentage of fillet weight have been reported to be 3.85 to 6.5 in salmonids (Kause et al., 2002Go; Neira et al., 2004Go; present study), 15.5 in tilapia (Rutten et al., 2005Go), and 5.3 in common carp (Kocour et al., 2007Go). Similarly, in our data the CV for fillet weight was lower than for BW and, in particular, for visceral weight. Weatherley and Gill (1983Go, 1987)Go found fillet percentage to be invariable across differently sized rainbow trout. In contrast to protein body percentage and fillet percentage, tissue and whole-body lipid percentages as well as visceral percentage are more prone to phenotypic and genetic variation, which makes selection for the lipid traits more effective (Tobin et al., 2006Go).

Relationships of BW, Fillet Traits, Viscera, and Head Volume

The phenotypic correlation between fillet percentage and BW was positive and favorable (0.22), but the genetic correlation was only weakly positive (0.04). In previous studies, this genetic correlation has been stronger (Kause et al., 2002Go: rP = 0.13, rG = 0.29; Neira et al., 2004Go: rP = 0.97, rG = 0.98; Rutten et al., 2005Go: rP = 0.48, rG = 0.74; Kocour et al., 2007Go: rP = 0.46, rG = 0.73). Eviscerated BW displayed a slightly more favorable correlation structure with fillet percentage compared with uneviscerated BW. Similarly, in our previous data the correlations of fillet percentage with eviscerated BW (rP = 0.23, rG = 0.47) were stronger than those with BW (rP = 0.13, rG = 0.29; Kause et al., 2002Go). Consequently, inclusion of eviscerated BW and not BW in the selection index is preferred when simultaneously improving growth and fillet percentage.

Visceral percentage, but not relative head volume, exhibited strong negative phenotypic and genetic correlations with fillet percentage. Similarly, Kause et al. (2002)Go and Kocour et al. (2007)Go found strong correlations between percentage of eviscerated BW and fillet percentage in rainbow trout (rP = 0.73, rG = 0.94) and common carp (rP = 0.63, rG = 0.79), respectively. Thus, selection against visceral percentage can be used to select indirectly for fillet percentage. In contrast to our estimate of –0.18, Rutten et al. (2005)Go found a very strong negative genetic correlation (–0.94) between fillet percentage and head weight percentage in Nile tilapia. In tilapia, the percentage of head weight is as high as 25% (Rutten et al., 2005Go), and this large value may explain these differences. In 2.5-kg Finnish rainbow trout, head weight accounts for 8.1% of the total BW (Kause et al., unpublished data).

Fillet weight, but not visceral weight and head volume, had close to unity phenotypic and genetic correlations with BW. Similar results for fillet and visceral weight were found by Kause et al. (2002)Go. This reveals, on one hand, severe constraints for changing fillet weight and BW independently by selection. On the other hand, this means that variation in BW predicts variation in fillet weight, which is beneficial for breeders. Consequently, sole selection for BW will be a very effective strategy to improve fillet weight.

Selection Strategy to Increase Fillet Percentage and Fillet Weight

Because of the high heritability of visceral percentage and its high genetic correlation with fillet percentage, indirect selection against visceral percentage was equally effective at increasing fillet percentage as direct selection for fillet percentage. Head dimensions can be recorded from live breeding candidates, but unfortunately, including relative head volume in the index increased selection accuracy only marginally. Thus, the ease of head dimension recording cannot be exploited for selection purposes. Both direct and indirect selection for increased fillet percentage led to a major genetic reduction in visceral weight and only a minor genetic improvement in fillet weight. This has 2 implications. First, we do not know how much the visceral percentage can be reduced without influencing fish health. Second, the major reduction in visceral weight will reduce production costs rather than increase economic return. For instance, it costs less to produce fillets in fish of the same weight that have a smaller percentage of viscera because of lower feed costs and facility requirements. Yet even a slight increase in fillet weight as a consequence of increased fillet percentage is likely to increase the economic return considerably.

Direct selection for fillet weight leads to higher genetic gain in fillet weight compared with indirect selection, but fillet weight recording is very laborious. Fortunately, selection for increased eviscerated BW led to parallel genetic changes in fillet weight. Thus, selection for increased eviscerated BW is a simple and cost-effective method of fillet weight improvement. Selection for increased eviscerated BW is also expected to lead to a small correlated genetic increase in fillet percentage. The correlated genetic increase in fillet percentage may, however, be underestimated in the current study, because the genetic correlation of fillet percentage with BW and eviscerated BW estimated here was lower than in previous studies (Kause et al., 2002Go; Neira et al., 2004Go; Rutten et al., 2005Go; Kocour et al., 2007Go).

Simultaneously selecting for eviscerated BW and against visceral percentage proved to be an effective indirect way of obtaining high genetic gain in fillet weight and of slightly improving fillet percentage. Both eviscerated BW and visceral percentage are easily measured during slaughter, and there is no need for laborious fillet weight recording. The disadvantage is that eviscerated BW and visceral percentage cannot be measured on live fish. Thus, eviscerated BW and visceral percentage need to be measured on the sibs of the breeding candidates. In the Finnish breeding program for rainbow trout, breeding candidates are held at the nucleus station under noncommercial conditions, whereas their sibs are reared and slaughtered at commercial fish farms at sea (Kause et al., 2005Go). During slaughter, fillet color is measured from the sea-reared sibs. Eviscerated BW and visceral percentage can easily be measured on these fish too, to genetically change visceral lipid, fillet percentage, and fillet weight.

In conclusion, the results suggest that fillet percentage can be efficiently improved by selection on the more easily measured visceral percentage. Likewise, fillet weight can be effectively improved by selection on BW or eviscerated BW. When destructively recording sibs of breeding candidates for slaughter and quality traits, eviscerated BW rather than wet BW should be used when simultaneously improving growth and visceral or fillet percentage.


   Footnotes
 
1 The authors acknowledge staff at the Tervo aquaculture station for fish management and trait recording, and Cheryl Quinton for comments on an earlier version of the manuscript. The study was funded by the Finnish Ministry of Agriculture and Forestry. Back

2 Corresponding author: Antti.Kause{at}mtt.fi

Received for publication June 7, 2007. Accepted for publication August 9, 2007.


   LITERATURE CITED
Top
 Abstract
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 LITERATURE CITED
 


Bergot, P., J. Blanc, and A. Escaffre. 1981. Relationship between number of pyloric caeca and growth in rainbow trout. Aquaculture 22:81–96.[CrossRef]

Bosworth, B. G., G. S. Libey, and D. R. Notter. 1998. Relationships among total weight, body shape, visceral components, and fillet traits in palmetto bass (striped bass female Morone saxatilis x white bass male M. chrysops) and paradise bass (striped bass female M. saxatilis x yellow bass male M. mississippiensis). J. World Aquac. Soc. 29:40–50.[CrossRef]

Bosworth, B. G., M. Holland, and B. L. Brazil. 2001. Evaluation of ultrasound imagery and body shape to predict carcass and fillet yield in farm-raised catfish. J. Anim. Sci. 79:1483–1490.[Abstract/Free Full Text]

Cameron, N. D. 1997. Selection Indices and Prediction of Genetic Merit in Animal Breeding. CAB Int., Wallingford, UK.

Cibert, C., Y. Fermon, D. Vallod, and F. J. Meunier. 1999. Morphological screening of carp Cyprinus carpio: Relationship between morphology and fillet yield. Aquat. Living Resour. 12:1–10.[CrossRef]

Elvingson, P., and K. Johansson. 1993. Genetic and environmental components of variation in body traits of rainbow trout (Oncorhynchus mykiss) in relation to age. Aquaculture 118:191–204.[CrossRef]

Elvingson, P., and J. Nilsson. 1994. Phenotypic and genetic parameters of body and compositional traits in Arctic charr, Salvelinus alpinus. Aquac. Fish. Manag. 25:677–685.

Falconer, D. S., and T. F. C. Mackay. 1996. Introduction to Quantitative Genetics. 4th ed. Longman Group Ltd., Essex, UK.

Gall, G. A. E. 1975. Genetics of reproduction in domesticated rainbow trout. J. Anim. Sci. 40:19–28.[Abstract/Free Full Text]

Gjedrem, T. 1997. Flesh quality improvement in fish through breeding. Aquacult. Int. 5:197–206.[CrossRef]

Gjerde, B., and L. R. Schaeffer. 1989. Body traits in rainbow trout. II. Estimates of heritabilities and of phenotypic and genetic correlations. Aquaculture 80:25–44.[CrossRef]

Hayes, J. F., and W. G. Hill. 1981. Modification of estimates of parameters in the construction of genetic selection indices (‘bending’). Biometrics 37:483–493.[CrossRef]

Hazel, L. N. 1943. The genetic basis of constructing selection indices. Genetics 28:476–490.[Free Full Text]

Iwamoto, R. N., J. M. Myers, and W. K. Hershberger. 1990. Heritability and genetic correlations for flesh coloration in pen-reared Coho salmon. Aquaculture 86:181–190.[CrossRef]

Jensen, J., and P. Madsen. 2000. A User’s Guide to DMU. A Package for Analyzing Multivariate Mixed Models. Natl. Inst. Anim. Sci., Res. Ctr., Foulum, Denmark.

Kause, A., O. Ritola, T. Paananen, E. Mäntysaari, and U. Eskelinen. 2002. Coupling body weight and its composition: A quantitative genetic analysis in rainbow trout. Aquaculture 211:65–79.[CrossRef]

Kause, A., O. Ritola, T. Paananen, H. Wahlroos, and E. A. Mäntysaari. 2005. Genetic trends in growth, sexual maturity and skeletal deformations, and rate of inbreeding in a breeding programme for rainbow trout. Aquaculture 247:177–187.[CrossRef]

Kause, A., D. Tobin, E. A. Mäntysaari, S. A. M. Martin, D. F. Houlihan, A. Kiessling, K. Rungruangsak-Torrissen, O. Ritola, and K. Ruohonen. 2007. Genetic potential for simultaneous selection of growth and body composition in rainbow trout (Oncorhynchus mykiss) depends on the dietary protein and lipid content: Phenotypic and genetic correlations on two diets. Aquaculture 271:162–172. .[CrossRef]

Kocour, M., S. Mauger, M. Rodina, D. Gela, O. Linhart, and M. Vandeputte. 2007. Heritability estimates for processing and quality traits in common carp (Cyprinus carpio L.) using a molecular pedigree. Aquaculture 270:43–50.[CrossRef]

Martinez, V., A. Kause, E. Mäntysaari, and A. Mäki-Tanila. 2006. The use of alternative breeding schemes to enhance genetic improvement in rainbow trout. II. Two-stage selection. Aquaculture 254:195–202.[CrossRef]

Neira, R., J. P. Lhorente, C. Araneda, N. Diaz, E. Bustos, and A. Alert. 2004. Studies on carcass quality traits in two populations of Coho salmon (Oncorhynchus kisutch): Phenotypic and genetic parameters. Aquaculture 241:117–131.[CrossRef]

Ouweltjes, W., L. R. Schaeffer, and B. W. Kennedy. 1988. Sensitivity of methods of variance component estimation to culling type of selection. J. Dairy Sci. 71:773–779.[Abstract/Free Full Text]

Poppe, T. T., R. Johansen, G. Gunnes, and B. Tørud. 2003. Heart morphology in wild and farmed Atlantic salmon Salmo salar and rainbow trout Oncorhynchus mykiss. Dis. Aquat. Org. 57:103–108.[Medline]

Rutten, M. J. M., H. Bovenhuis, and H. Komen. 2004. Modeling fillet traits based on body measurements in three Nile tilapia strains (Oreochromis niloticus L.). Aquaculture 231:113–122.[CrossRef]

Rutten, M. J. M., H. Bovenhuis, and H. Komen. 2005. Genetic parameters for fillet traits and body measurements in Nile tilapia (Oreochromis niloticus L.). Aquaculture 246:101–113.[CrossRef]

Rye, M., and B. Gjerde. 1996. Phenotypic and genetic parameters of body composition traits and flesh colour in Atlantic salmon Salmo salar L. Aquac. Res. 27:121–133.

Shearer, K. D. 1994. Factors affecting the proximate composition of cultured fishes with emphasis on salmonids. Aquaculture 119:63–88.[CrossRef]

Sokal, R. R., and F. J. Rohlf. 1981. Biometry. W. H. Freeman, New York, NY.

Su, G. S., L. E. Liljedahl, and G. A. E. Gall. 1997. Genetic and environmental variation of female reproductive traits in rainbow trout (Oncorhynchus mykiss). Aquaculture 154:115–124.[CrossRef]

Tobin, D., A. Kause, E. A. Mäntysaari, S. A. M. Martin, D. F. Houlihan, A. Dobly, A. Kiessling, K. Rungruangsak-Torrisen, O. Ritola, and K. Ruohonen. 2006. Fat or lean? The quantitative genetic basis for selection strategies of muscle and body composition trait in breeding schemes of rainbow trout (Oncorhynchus mykiss). Aquaculture 261:510–521.[CrossRef]

Weatherley, A. H., and H. S. Gill. 1983. Relative growth of tissues at different somatic growth rates in rainbow trout Salmo gairdneri. J. Fish Biol. 22:43–60.[CrossRef]

Weatherley, A. H., and H. S. Gill. 1987. The Biology of Fish Growth. Academic Press, London, UK.

Wray, N. R., and M. E. Goddard. 1994. Increasing long-term response to selection. Genet. Sel. Evol. 26:431–451.[CrossRef]



This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow All Versions of this Article:
jas.2007-0332v1
85/12/3218    most recent
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Kause, A.
Right arrow Articles by Koskinen, H.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Kause, A.
Right arrow Articles by Koskinen, H.

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS